A scoping review was conducted in accordance with the PRISMA-ScR (Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews) guidelines [13] to identify studies that examined patients/consumers, providers, and/or population stakeholders in primary care settings (eg, limited to outpatient, ambulatory care, and long-term care) that used at least one digital health technology as an intervention for prevention (primary [eg, timely receipt of vaccinations], secondary [eg, completion of indicated screenings], tertiary [eg, routine monitoring of chronic conditions], and quaternary [eg, prevention of overmedicalization]) and reported beneficial outcomes on health, health care performance, and implementation science. The protocol is available upon request.

Systematic search queries of MEDLINE via PubMed, Embase, and the Cochrane Library were used to identify references published or available online between January 1, 2014, and July 19, 2020 (Multimedia Appendices 1-7). Studies were limited to primary designs or systematic reviews (with the same inclusion criteria) published in English with abstracts. The rationale for this search cutoff time frame was based upon a high threshold of eligible providers achieving meaningful use of certified electronic health record (EHR) technology, whereby 82.8% of office-based physicians had adopted any EHR [14].

To ensure screener alignment, dual review of 20% of randomized titles and abstracts was followed by group resolution of conflicts. All remaining titles and abstracts underwent single review, and full-text articles were examined by 2 independent reviewers for relevance against the inclusion/exclusion criteria (Multimedia Appendix 8), with third-party adjudication provided for any discrepancies in eligibility. Results were tracked in DistillerSR (Evidence Partners).

The eligible population criteria included studies that examined patients/consumers, providers (both licensed and unlicensed), and/or population stakeholders (eg, payers, employers, communities, health systems, and the government) in outpatient care, ambulatory care, and long-term settings of primary care. Interventions had to target primary, secondary, tertiary, or quaternary prevention using at least one FDA/WHO approved or nonregulated digital health technology facet (eg, telehealth, mHealth, HIT, data analytics, and AI). No comparisons were required. Outcomes of interest included health (eg, individual- or population-level outcomes), health care performance (eg, as per the Agency for Healthcare Research and Quality [AHRQ]: access, quality, utilization, and efficiency, with measures categorized as structural, process, or outcomes including clinical/physiological, surrogate/intermediate, patient-centered, or patient-reported), and DHI implementation (eg, taxonomy as per Proctor et al: acceptability, adoption, appropriateness, costs, feasibility, fidelity, penetration, and sustainability) [15]. Only English-language primary studies or systematic reviews with the same inclusion criteria published between January 2014 and July 2020 were included. For definitions and descriptions of terms, see Multimedia Appendix 8 and Multimedia Appendix 9. Notable exclusion criteria for interventions included DHIs associated with treatment or diagnosis (except for preventive screenings), medical imaging for diagnosis, and telehealth using only noncellular telephone communication. Studies conducted in critical care (eg, intensive care unit) or inpatient (eg, hospital admission) settings were excluded.

After a series of data form piloting and discussions by all extractors to identify gaps in data extraction forms and ensure consistency in the application of definitions, data were abstracted into standardized forms within DistillerSR (Multimedia Appendix 10) for synthesis by a single reviewer. All fields of the data extraction forms for each article were examined for completeness by a second reviewer. Many data categorizations were not mutually exclusive, resulting in percentages totaling more than 100%.

Following title and abstract screening, the large scope (>1000 titles) of the remaining included studies prohibited full-text review of all preventive DHIs identified globally. To narrow the scope of the geography and interventions under review, a subgroup analysis was performed; geography limits were set to only include studies conducted in the US. Additionally, it was apparent that a large volume of records focused on data analysis methods tangential to the development of DHIs. As such, studies that only used EHRs as a retrospective data capture tool were excluded. Two examples of excluded studies are a retrospective analysis of EHRs to determine the prevalence of a preventable disease and a study on the use of diagnostic telemedicine referral to a dermatologist.

Content analysis of extracted technology descriptions was performed to identify recurrent topics and more clearly understand the types of DHIs evaluated in the included studies according to a priori research questions in the protocol. This analysis yielded a list of articles selected to represent innovative or unique DHIs and their implementation in the final data set. Selected technologies were then narratively synthesized into 5 topical groups (eg, population-centered, patient-centered, care access expansion, panel-centered [dashboarding], and app-driven) to provide a framework for their analysis. Selected outcome (eg, health, health care performance, and implementation science) results from these articles were then extracted by a single reviewer to provide additional context regarding the impact of these DHIs beyond the directionality of their results. Details presented from this synthesis are not exhaustive, and key use cases have been highlighted in the results.

Study quality was assessed using the Oxford levels of evidence [16], which allow for the categorization of evidence quality across heterogeneous study types. Examples of the study types comprising these evidence levels include (in increasing quality) expert opinion, case series, systematic reviews of case-control studies, individual cohort studies, randomized controlled trials (RCTs) with narrow confidence intervals, and systematic reviews of RCTs.

Amidst the rapid digital transformation of the primary care delivery system in response to the COVID-19 pandemic, this is the first comprehensive summary on DHIs in use by interdisciplinary clinicians (eg, physicians, pharmacists, psychiatrists, etc) in primary care. This scoping review and its subgroup analysis summarized a growing evidence base and rendered a collection of potentially successful strategies for patients, providers, and population stakeholders to improve outcomes for health, health care performance, and implementation science through the use of DHIs. Moreover, important scientific gaps were identified in the contemporary evaluation and knowledge of DHIs leveraged in primary care, particularly the scarcity of the evaluation of DHIs in health disparities and evaluation of the negative effects of DHIs.

A few major themes emerged from our analysis of the extracted data. First, the digital health technologies identified and reviewed were highly concentrated in a narrow range of HIT, most specifically around EHRs/electronic medical records, particularly with the use of alerts to help clinicians make appropriate clinical decisions. Though understandable given their high use and decade-long attention to increasing adoption via “meaningful use” in primary care [259,260], the absence of DHI literature involving other platforms was telling. Despite unprecedented attention to telehealth implementation due to the COVID-19 pandemic response, little evidence of effective implementation of this specific DHI exists to guide primary care telehealth use for health care delivery in the US. A few studies did examine more innovative uses of technology, particularly for the delivery of mental and behavioral health (Tables 1-5). As HIT continues to rapidly evolve and health care is delivered in more innovative ways due to the COVID-19 pandemic, more research should focus on novel DHIs applied to primary care.

Second, despite prevention being 1 of 6 mechanisms underpinning primary care’s beneficial impact on population health [261] and an early target for DHIs, studies evaluating prevention were predominantly focused on secondary or tertiary preventive interventions. Most would agree that disease prevention offers the greatest yield for population health and is amenable to DHIs via mobile and online apps, clinical kiosks, and electronic patient portals [262,263]. Primary prevention interventions, such as immunizations, rely on effective patient counseling and education, which can be difficult and time-consuming to document and capture in EHRs (the predominant type of intervention found in our review). This finding may be a reflection of physician roles in the US. Traditionally, the role of primary prevention has relied on public health professionals [264], and although primary care physicians are increasing their ability to address the needs of the community, most physicians are still focused on the needs of the individual [265]. As the intersection of public health and primary care becomes more urgent to strategically improve individual and population health, future studies should examine the role of DHI adoption and implementation in their integration.

Third, DHIs enhanced core primary care functions by contributing to the comprehensiveness of care provided. This was an unexpected finding given that DHIs are often thought of in the context of first contact through patient portals, coordination through electronic referrals and linked EHRs, and continuity through HIEs and sharing of documents. Many of the articles reviewed discussed the use of DHIs to identify patients in need of services and alert clinicians to provide them. For example, multiple studies described EHR alerts that would prompt clinicians to order viral hepatitis C testing for patients with indications for screening (Multimedia Appendix 11) [83,93,112,116,133,150,164,173,207,246]. Other studies shared examples of how patients could be trained to provide services for themselves (Table 2) or how DHIs could be used to offer additional clinical services (Table 3). Thus, it makes sense that comprehensiveness, or the provision of a robust set of services to a patient, would be improved with DHIs. In an era where comprehensiveness of care is said to be declining in primary care [266-271], DHIs may provide an innovative solution for primary care practices to increase and enhance the services they provide.

Finally, while the development and release of health apps continue to increase, few evaluations of app-driven DHIs were identified in our study (Table 5). This may be in part because many apps lacked integration with primary care or other technology systems or because of the evolving standards for evaluating these types of interventions, as evidenced by the recent establishment of the FDA’s Digital Health Center of Excellence [272]. Most app-driven DHIs included in our study were patient-facing and focused on helping to better involve patients in their care. However, app-driven DHIs are also capable of providing an overwhelming amount of data to providers. Balancing data collection features from apps by adding functionalities, such as thresholds triggering clinical alerts/feedback, designing patient-counseling suggestions based on gathered data, and pairing with timely coaching/contact is important to enhance the clinical relevance and quality of these tools. As the development and clinical adoption of app-driven DHIs continue to expand, rigorous investigation of their safety, efficacy, and value in primary care is urgently needed.

These results should be interpreted in the context of a few limitations. The findings are limited to studies conducted in US settings, which prohibits the generalization of their applicability and use at a global scale. Review of the use of DHIs in non-US primary care settings should be prioritized in future work. Further, due to the heterogeneity of identified interventions, it is not possible to provide head-to-head comparisons. The large heterogeneity of DHIs is an additional reason why our synthesis focused on the novel and distinct DHIs that are collectively used in primary care practice rather than presenting evidence collated by distinct DHI technologies. Other limitations include single screening of titles and abstracts, English language restriction, and lack of gray literature evaluation. Data extraction for each article was not confirmed by a secondary reviewer, leaving room for bias in the interpretation of the articles. For example, it was left up to each reviewer to determine the type of prevention the DHI was addressing, or which primary care function (eg, the 4Cs by Dr Starfield: first contact, comprehensiveness, continuity, and coordination) the DHI enhanced. However, careful and collaborative definition of our processes and outcomes prior to extraction (ie, types of prevention or primary care functions) should minimize this bias. Lastly, we intentionally selected a quality assessment tool rather than a risk of bias tool, as we only planned to measure the extent that methodological safeguards (ie, internal validity) against bias were implemented. A risk of bias assessment would have offered a bias judgement (ie, estimation of intervention effects) on such a quality assessment, and judgement of the evidence may have shifted with this approach. It is important to consider that even when a study implements all possible safeguards in a tool, it may not be unbiased, and conversely, a study applying no safeguards is not necessarily biased [273].

Gayle Stephens noted in 1965 that “One of the paradoxes of our time is that the healing relationship seems most in jeopardy at a time when we need it most,” commenting on the range of “forces which threaten to depersonalize the meeting of a doctor and patient” [274]. That paradox remains in an age where technology is often seen as distracting rather than enhancing care. Through further adoption of DHIs with evidence of effectiveness, providers and patients/consumers can enhance primary care by improving the delivery of preventive services and promoting more comprehensive care. Yet, relying solely on EHR alerts may not lead to substantial improvements in health care in the US. Moreover, rigorous and prospective evaluations of the potential negative effects of these DHIs, particularly for clinical end users of these technologies, will be needed to ensure holistic improvement of health care. Innovative DHIs should undergo evaluation in well-designed studies to generate evidence and establish best practices that can be replicated and scaled in diverse primary care settings. Given the ability of technology to amplify existing health disparities and biases, the development of DHIs that can help overcome health disparities and the evaluation of the benefits and harms of current DHIs on health disparities are imperative. In addition, DHIs that allow integration of public health with primary care will be essential for rapid and effective responses to health and health care challenges, such as the COVID-19 pandemic, in an increasingly technology-driven health care environment.